Replicate
Replicate runs AI models via cloud API and provides model execution, fine-tuning, and custom deployment paths.
Who it is for: Developers, researchers, enterprises
Core capabilities
- Cloud API Model Execution: Run AI models via cloud API.
- Model Fine-tuning: Provides an introductory path for image model fine-tuning and claims to support model fine-tuning.
- Custom Model Deployment: Package machine learning models with Cog and deploy them to cloud clusters.
- Auto Scaling: The official site claims the platform automatically scales up and scales down to zero without charge when there's no traffic.
- Runtime Observability: Provides descriptions related to prediction throughput, metrics, and logs for runtime observability.
Pricing
Public models are typically billed by runtime and hardware; some are billed by input and output. Most private models are billed by instance uptime. The official website provides a free start entry, but the free quota is not clearly defined.
Updated 2026-08-18